Cryptic Species Discovered in Biological Control Agent Populations Using Molecular Data and Crossbreeding Experiments Iain D

Two in one: cryptic species discovered in biological control agent populations using molecular data and crossbreeding experiments Iain D. Paterson1, Rosie Mangan1, Douglas A. Downie1, Julie A. Coetzee1, Martin P. Hill1, Ashley M. Burke1, Paul O. Downey1,2, Thomas J. Henry3 & Stephe G. Compton1 1Department of Zoology and Entomology, Rhodes University, PO Box 94, Grahamstown 6140, South Africa 2Institute for Applied Ecology, University of Canberra, Canberra, Australian Capital Territory 2601, Australia 3Systematic Entomology Laboratory, ARS, USDA, c/o National Museum of Natural History, Smithsonian Institution, Washington, District of Columbia 20013

Keywords Abstract Biological species concept, C01, DNA- barcoding, Eccritotarsus catarinensis, There are many examples of cryptic species that have been identified through Eichhornia crassipes, Inter Simple Sequence DNA-barcoding or other genetic techniques. There are, however, very few con- Repeats, reproductive incompatibility. firmations of cryptic species being reproductively isolated. This study presents one of the few cases of cryptic species that has been confirmed to be reproduc- Correspondence tively isolated and therefore true species according to the biological species con- Iain D. Paterson, Department of Zoology and cept. The cryptic species are of special interest because they were discovered Entomology, Rhodes University, PO Box 94, within biological control agent populations. Two geographically isolated popu- Grahamstown 6140, South Africa. Tel: +27 46 603 8098; lations of Eccritotarsus catarinensis (Carvalho) [Hemiptera: Miridae], a biologi- Fax: +27 46 622 8959; cal control agent for the invasive aquatic macrophyte, water hyacinth, E-mail: [email protected] Eichhornia crassipes (Mart.) Solms [Pontederiaceae], in South Africa, were sampled from the native range of the species in South America. Morphological Funding Information characteristics indicated that both populations were the same species according South African Agency for Science and to the current taxonomy, but subsequent DNA analysis and breeding experi- Technology Advancement (Grant/Award Number: 84643), Working for Water (DEA: ments revealed that the two populations are reproductively isolated. Cross- NRM). breeding experiments resulted in very few hybrid offspring when individuals were forced to interbreed with individuals of the other population, and no Received: 29 January 2016; Revised: 10 June hybrid offspring were recorded when a choice of mate from either population 2016; Accepted: 10 June 2016 was offered. The data indicate that the two populations are cryptic species that are reproductively incompatible. Subtle but reliable diagnostic characteristics Ecology and Evolution 2016; 6(17): 6139– were then identified to distinguish between the two species which would have 6150 been considered intraspecific variation without the data from the genetics and doi: 10.1002/ece3.2297 interbreeding experiments. These findings suggest that all consignments of biological control agents from allopatric populations should be screened for cryptic species using genetic techniques and that the importation of multiple consignments of the same species for biological control should be conducted with caution.

Introduction are usually used to describe species (Schlick-Steiner et al. 2007; Cook et al. 2010). The result of this incongruence The biological species concept defines species as inter- between how species are defined and how new species are breeding populations that are reproductively isolated from described is that many described species may comprise other populations (Mayr 1942, 1963; Noor 2002). multiple biological species that appear to be morphologi- Although the biological species concept is generally cally indistinguishable or identical. Two or more species accepted by the majority of evolutionary biologists (Noor that have been classified as a single species due to their 2002; Coyne and Orr 2004), morphological characteristics, morphological characteristics are referred to as cryptic rather than the ability of biological entities to interbreed, species (Darlington 1940).

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 6139 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Cryptic Species in Biological Control I. D. Paterson et al.

There is growing evidence that cryptic species are much negative consequences of a genetic bottleneck can be more common than was previously thought (Hebert et al. overcome by combining individuals from different source 2004a; Bickford et al. 2007). The growth in the number populations (Colautti et al. 2005; Lachmuth et al. 2010) of recorded cryptic species is primarily due to the use of so it was expected that when the two populations were DNA-barcoding which can successfully distinguish allowed to interbreed, the resulting population would be between what appear to be morphologically identical spe- a more effective agent (Taylor et al. 2011). cies based on the sequence divergence within and between The Brazilian population, which was first released in species (Hebert et al. 2003). While the occurrence of South Africa in 1996, has established widely throughout many potential cryptic species have been identified using the country and contributes toward the control of water DNA-barcoding (e.g., Wilcox et al. 1997; Gomez et al. hyacinth, especially during periodic outbreaks of the agent 2002; Hebert et al. 2004a,b; Miura et al. 2005; Ward et al. that result in the collapse of water hyacinth mats at some 2008; Saitoh et al. 2015), few have been confirmed to be sites (Winston et al. 2014). Although biological control reproductively incompatible using crossbreeding experi- has successfully reduced the water hyacinth problem in ments in combination with molecular data. For some South Africa, complete control has not been achieved organisms, it is not possible to conduct crossbreeding (Coetzee et al. 2011). Cold temperatures (Coetzee et al. experiments, so multiple data sets combining morpholog- 2007a), as well as high nutrients in South African aquatic ical, ecological, and physiological data have been com- ecosystems (Coetzee et al. 2007b), are possible limiting bined with genetic data to provide a better understanding factors to the efficacy of E. catarinensis. The low genetic of whether true biological species are present (e.g., Ross diversity of the Brazilian population has been regarded as and Shoemaker 2005; Rissler and Apodaca 2007; Abad- another factor that could be limiting success, so the more Franch et al. 2013; Davis et al. 2016; Dejaco et al. In genetically diverse Peruvian population was released in press). Although these approaches are valuable, inter- 2007 with the intention of increasing the level of control breeding experiments are a more direct and conclusive provided by the agent (Taylor et al. 2011). Subsequent method of testing for reproductive incompatibility. There releases have been conducted but establishment of the are very few examples of cryptic species being verified in Peruvian population is yet to be confirmed (Winston this way. Some examples include the sweet potato white- et al. 2014). The interactions between the two popula- fly, Bemisia tabaci (Gannadius) [Hemiptera: Aleyrodidae] tions, and how each population performs at different (Xu et al. 2010; Liu et al. 2012), the dipteran vector of temperatures, is likely to play an important role in the the disease leishmaniasis, Lutzomyia longipalpis Lutz and biological control programme against water hyacinth in Neiva [Dipetra: Psychodidae] (Souza et al. 2008), and the South Africa and elsewhere in the world where E. catari- biological control agent for cereal stem borers, Cotesia nensis may be released in future (Taylor et al. 2011). sesamiae (Cameron) [Hymenoptera: Braconidae] (Kaiser A comparison of the mitochondrial cytochrome oxidase et al. 2015). This paper presents the first example of cryp- subunit 1 gene (COI) of the Peruvian and Brazilian popu- tic species confirmed as reproductively incompatible taxa lations imported to South Africa indicated a surprisingly nested within weed biological control agent populations. large sequence divergence of 5.2% between the popula- Eccritotarsus catarinensis (Carvalho) [Hemiptera: Miri- tions, suggesting the possibility that they should be con- dae] is a natural enemy of water hyacinth (Eichhornia sidered cryptic species (Taylor et al. 2011). Inter Simple crassipes (Mart.) Solms. [Pontederiaceae]) used for biolog- Sequence Repeat (ISSR) data supported this conclusion as ical control in South Africa (Coetzee et al. 2011). The 29 fixed differences were observed (Taylor et al. 2011). agent was collected in its native range of South America Sequence divergences at the COI region have been investi- from two geographically isolated populations that are sep- gated in 16 other species of Miridae and only one species arated by a distance of over 3500 km (Taylor et al. 2011). had a mean divergence value higher than 5.2%, whereas The first collection of the agent for biological control was four of the species had a maximum intraspecific diver- made in Florianopolis (Santa Catarina), Brazil in 1994, gence distance greater than this value (Park et al. 2011). and this collection was subjected to host specificity testing These data indicate that most Miridae have lower and then released throughout South Africa (Hill et al. intraspecific sequence divergence than that found between 1999). Prior to release, the Brazilian collection had been the Peruvian and Brazilian populations of E. catarinensis subjected to a severe genetic bottleneck event after impor- but also that there are some exceptions. Therefore, the tation into quarantine so a subsequent collection was presence of cryptic species in E. catarinensis cannot be made on the Yarapa River near Iquitos, Peru in 1999 solely inferred by the high sequence divergence alone. (Cordo 1999), with the intention of interbreeding the The aim of this study was to test the hypothesis that populations for the purpose of increasing genetic diversity the two populations of E. catarinensis are cryptic species (Taylor et al. 2011). There is strong evidence that the according to the biological species concept.

6140 ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. I. D. Paterson et al. Cryptic Species in Biological Control

Morphological characters of the two populations were left paramere and the width of the hook at the end of the compared to confirm the morphological similarity of the left paramere were measured. Data for all morphological two populations. The ability of the two populations to parameters were normally distributed so the populations interbreed was then tested under no-choice and choice were compared using t-tests in STATISTICA ver. 13 (Stat- conditions. Under no-choice conditions, the insects could Soft Inc., Tulsa, OK). either attempt to reproduce with insects from the other population or not reproduce at all. Under choice condi- No-choice interbreeding experiment tions, the insects were given a choice of reproducing with mates from either population. Late-instar nymphs were collected from each of the populations and placed individually in Petri dishes with damp- Methods ened filter paper and a water hyacinth leaf-disk approximately 2 cm in diameter. The Petri dishes were labeled according to the population source of the nymph Morphological comparison and sealed with Parafilm to maintain humidity. Leaf- Eccritotarsus catarinensis was described from a holotype disks and moist filter paper were replaced when necessary. male and four male and nine female paratypes (including Keeping the nymphs individually until their final ecdysis an allotype) collected in the state of Santa Catarina in to the adult stage ensured that all the individuals used in southern Brazil (Carvalho 1948). In 2012, small series of this experiment were unmated. both sexes from each E. catarinensis population were sent Within 12 h of the final ecdysis, adults were removed to the Systematic Entomology Laboratory ARS-USDA, in from the Petri dishes and divided into breeding pairs. Washington, DC, for identification by Thomas J. Henry, The four breeding pair treatments were: a specialist on the family Miridae. This was performed to Brazil♀ 9 Brazil♂ (n = 10); Peru♀ 9 Peru♂ (n = 15); confirm that both populations should be considered as Brazil♀ 9 Peru♂ (n = 27) and Peru♀ 9 Brazil♂ (n = 24). E. catarinensis according to the current taxonomic knowl- Each breeding pair was placed on a small water hyacinth edge of the group (Carvalho 1948). Based on comparison plant in a clear plastic jar half filled with water and cov- with six paratypes in the United States National Museum ered with nylon mesh. The plants were examined daily of Natural History, Washington, DC, the series were con- for the presence of nymphs for a 30-day period and the firmed to be E. catarinensis. Subsequently, the specialist total number of nymphs produced by each breeding pair was made aware that the two series could represent cryp- was recorded. tic species due to high levels of sequence divergence in Data were not normally distributed so a Kruskal–Wallis the COI region and the large differences in the ISSR data ANOVA and a Multiple Comparison of Mean Ranks were (Taylor et al. 2011). Larger series of both populations conducted in STATISTICA v. 13 (StatSoft Inc. 1984– were then sent for examination with the intention of 2014). Hybrid offspring and the parents of hybrids were identifying reliable morphological characteristics to sepa- collected and preserved individually in 100% ethanol for rate the two populations. genetic analysis to verify that they were true hybrids. Morphological parameters of specimens from the two Genetic analysis was conducted according to the methods populations were also compared with each other and with of Taylor et al. (2011) using a combination of the COI the original description of the species. The morphological mitochondrial sequence and the nuclear ISSR profile to parameters provided for the type specimen in the descrip- determine whether offspring were Brazilian, Peruvian or tion include body length, body width, the parameters of hybrids (see Genetic analysis below for more details of the head (height, width, length), the length of each of the genetic analysis). four antennal segments, and the length and width of the pronotum (Carvalho 1948). The same measurements were Choice interbreeding experiment taken for ten males and ten females from each populations using an Olympus 52X16 compound microscope Choice tests were included because no-choice situations and the imaging software programme Olympus Stream are unlikely to occur under natural conditions and Motion (Olympus Corporation Ltd, Tokyo, Japan). The because no-choice tests may produce false positive results clasper organs, or parameres, of male Eccritotarsus are (Liu et al. 2012). important diagnostic features that are used to distinguish For choice experiments, eight 70-L tubs filled with water between the various species within the genus (Carvalho hyacinth and covered with a fine (0.5 mm) mesh net were 1948). For this reason, the parameres were dissected from inoculated with 200 E. catarinensis individuals. Two of the ten males of each population, mounted on microscope tubs were controls, one with 200 Brazilian E. catarinensis slides and examined in greater detail. The length of the and the other with 200 Peruvian E. catarinensis. The other

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 6141 Cryptic Species in Biological Control I. D. Paterson et al. six tubs (referred to as replicates 1–6) were founded with COI haplotype for that sample was used to identify whether 100 individuals from each population. The experiment was an individual was a pure Brazilian, pure Peruvian or hybrid. conducted in a greenhouse at Rhodes University, Graham- The mother’s population of a hybrid offspring was identified stown, South Africa, for 6 months from February to from the COI haplotype of the nymph. August 2011. After the 6-month period, nine individuals from each of the tubs were randomly sampled and stored Screening for Wolbachia individually in 100% ethanol for genetic analysis. Adult E. catarinensis live for an average of about 50 days (Hill Wolbachia, a bacterial endoparasite that is common in et al. 1999), so all the insects used at the start of the exper- arthropods, can result in cytoplasmic incompatibility iment would have been dead by the time that sampling between intraspecific lineages and could therefore be a took place. The insects sampled for genetic analysis were confounding factor in interbreeding experiments to con- therefore the progeny of those used to start the experi- firm reproductive isolation (Bordenstein et al. 2001). For ment, not the same individuals. Extractions were per- example, cytoplasmic incompatibility has been reported formed on nine individuals from each replicate and the between infected and uninfected lineages of the predatory two controls but in some cases, extractions or PCRs failed. mirid, Macrolophus pygmaeus (Rambir), resulting in a Sixty-eight individuals were sampled in total, including reduction in the number of offspring produced when nine from each of the two controls. crossing infected males with uninfected females but not when uninfected males were crossed with infected females (Machtelinckx et al. 2009). Wolbachia has also been Genetic analysis shown to promote speciation through bidirectional cyto- Taylor et al. (2011) characterized the Peruvian and Brazil- plasmic incompatibility between closely related species of ian populations using the COI region and ISSR profiles. parasitic wasps (Bordenstein et al. 2001). If one of the Following from their results, if hybrid individuals were E. catarinensis populations had a high rate of Wolbachia sampled, they would be expected to be intermediate in a infection, or if both populations had distinct Wolbachia nonmetric Multi-Dimensional Scaling (nMDS) scatterplot strains, then it is possible that interbreeding experiments from the nuclear (ISSR) data, and have an identical COI could falsely identify the two populations as cryptic spe- (mitochondrial) sequence to their mother. cies because crosses between infected and uninfected indi- A whole E. catarinensis adult or nymph was used for viduals, or individuals infected with different Wolbachia each DNA extraction. DNA was extracted, the COI region strains, may not produce offspring. Both E. catarinensis was sequenced and ISSR analysis was performed accord- populations were therefore screened for the presence of ing to the methods outlined in Taylor et al. (2011). The Wolbachia. COI region was amplified using the LCO 1490 and HCO Wolbachia positive controls were sourced from an 2198 primers (Hebert et al. 2003). ISSRs were conducted infected culture of Eldana saccharina Walker (Pyralidae) using the universal primers 809 and 826 from University housed at the South African Sugar Research Association of British Columbia Nucleic Acid Protein Service Unit (SASRI), Mount Edgecombe, KwaZulu-Natal, South Africa. Primer set #9 (Abbot 2001), fluorescently labeled with 6- The wsp 81F and 69IR primers were selected for Wolbachia FAM dye. Electropherograms were analyzed using PeakS- screening. These primers amplify a 590–632 bp fragment of canner v. 1.0 (Applied Biosystems, Foster City, CA) using the wsp gene (Zhou et al. 1998). The wsp primers amplify a medium level of peak smoothing. The analyzed data set most Wolbachia strains and produced visible bands for all was then exported into RawGeno v. 2.0-2 (Arrigo et al. Wolbachia infected arthropod species tested in a recent 2009) using the R© v. 3.0.1 platform (The R Foundation assessment of methods for screening for Wolbachia (Simoes~ for Statistical Computing, Vienna, Austria). Every sample et al. 2011). Extractions were conducted as reported above was replicated from the PCR step so there were two sets for the E. saccharina control and twenty E. catarinensis of binary data for each sample. Only peaks that were pre- from each population. Each 20 lL reaction contained sent in both replicates were used for further analyses. 10 lL of GoTaq Colorless Master Mix, 3 lL of extracted Bands that were only present in one of the replicates were DNA, 0.4 lL of each primer, 5.8 lL of nuclease-free water treated as missing data. The final binary data set was then and 0.4 lL of MgCl to give a final MgCl concentration of used to conduct an MDS analysis in PAST: Paleontologi- 2.5 mmol/L. The PCR protocol had an initial denaturing cal Statistics Package v. 1.94 (Hammer and Harper 2001) step of 94°C for 1 min followed by 32 cycles of 94°C for using Jaccard’s Index to convert the binary data into a 1 min, 50°C for 45 sec, 72°C for 1 min and a final exten- distance matrix. sion of 4 min at 72°C. The PCR products were then run for The position of each sample on the nMDS scatterplot 30 min at 80 V on a 1.2% agarose gel stained with SybrSa- compiled using the ISSR data, along with the corresponding feTM (Applied Biosystems, Foster City, CA) to visualize

6142 ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. I. D. Paterson et al. Cryptic Species in Biological Control bands under ultraviolet light. A 100 bp DNA ladder was Three of the 13 morphological features that were mea- included to estimate the size of any fragments that were sured were significantly different between the two popula- produced. tions for male specimens (Table 1) and four were significantly different for female specimens (Table 2). The Results width of Brazilian males was slightly greater than that of Peruvian males for both body width as well as the width at the pronotum, but the average difference was less than Morphological comparison 0.1 mm and overlaps in the size ranges for each popula- As noted, the Brazilian and Peruvian series of specimens tion did occur (Table 1). The fourth antennal segment of were all identified originally as E. catarinensis (Carvalho) Brazilian males was also 0.1 mm longer than that of by mirid specialist T.J. Henry (Systematic Entomology Peruvian males but again overlaps in the size ranges Laboratory, Agricultural Research Service, United States of the two populations occurred (Table 1). No significant Department of Agriculture, c/o National Museum of Nat- differences in the length of the paramere or the hook on ural History, Smithsonian Institution, Washington, DC). the paramere were detected (Table 1). Brazilian females These specimens are housed at the South African were significantly larger in length, width, pronotum width National Collection of Insects (A.R.C.), Pretoria, under and head width, but the ranges of all parameters over- the accession numbers AcP 9552 (Peru) and AcP 9553 lapped with the exception of width, and the ranges of this (Brazil), and in the United States National Museum of parameter were as close as 0.01 mm (Table 2). The Natural History, Washington, DC. Re-examination of the lengths of the different body parts are therefore unlikely original series and subsequent study of a much larger to be an accurate and consistent method to differentiate sample of specimens from each locality revealed a number between the two taxa. Other morphological features, such of specific differences between the Peruvian and Brazilian as pubescence and puncturing of the pronotum as well as specimens, including the width of the pronotum, the size the shape of the male genetalia (see above), will need to and density of punctures and pubescence on the prono- be included in combination with body size to distinguish tum; and male genitalia, especially the width of the basal between the two populations. Differences between the lobe on the left paramere and the shape of the middle measurements of the type specimen and those taken in process on the right paramere. These characters were this study can be attributed to the better accuracy with found to consistently distinguish the two taxa, but the which the parameters were measured due to the more differences were subtle and without having had the repro- sophisticated microscopes and measuring programmes ductive incompatibility evidence and the information that are now available. indicating large sequence divergences, these differences would have been considered interpopulation-level mor- No-choice interbreeding experiment phological variation. Hybrid nymphs were used for genetic analyses before they developed to the adult stage Fourteen of the 15 replicates (93%) of Peru♀ 9 Peru♂ so no morphological analysis of hybrids was possible. crosses produced offspring and 8 of 10 (80%) of the

Table 1. Morphological measurements of male Eccritotarsus catarinensis from two source populations, as well as the measurements of the type specimen (Carvalho 1948).

Parameter Carvalho (1948), mm Brazil (ÆSE), mm Peru (ÆSE), mm Brazil range, mm Peru range, mm t Value P value

Body length 3 2.87 (Æ0.063) 2.72 (Æ0.079) 2.68–3.36 2.52–3.39 À1.5 0.157 Body width 0.7 0.72 (Æ0.011) 0.63 (Æ0.009) 0.64–0.75 0.58–0.66 À6.3 <0.001 Pronotum length 0.7 0.72 (Æ0.007) 0.67 (Æ0.008) 0.64–0.84 0.61–0.74 À1.9 0.067 Pronotum width 1.3 0.81 (Æ0.011) 0.73 (Æ0.016) 0.76–0.84 0.69–0.77 À5.0 <0.001 Head length 0.2 0.23 (Æ0.006) 0.23 (Æ0.008) 0.20–0.28 0.20–0.27 0.1 0.978 Head width 0.5 0.58 (Æ0.005) 0.56 (Æ0.010) 0.55–0.68 0.52–0.69 À1.1 0.296 Head vertex 0.31 0.30 (Æ0.011) 0.31 (Æ0.011) 0.28–0.34 0.29–0.37 1.8 0.084 Antennae Segment 1 0.3 0.30 (Æ0.016) 0.29 (Æ0.015) 0.29–0.33 0.25–0.37 À1.2 0.265 Antennae Segment 2 0.5 0.55 (Æ0.015) 0.52 (Æ0.014) 0.47–0.59 0.46–0.58 À1.2 0.135 Antennae Segment 3 0.7 0.71 (Æ0.022) 0.69 (Æ0.014) 0.60–0.79 0.62–0.79 À0.9 0.375 Antennae Segment 4 0.6 0.64 (Æ0.012) 0.54 (Æ0.008) 0.57–0.70 0.50–0.65 À3.6 0.002 Length of left paramere Not measured 0.22 (Æ0.041) 0.21 (Æ0.055) 0.20–0.24 0.18–0.22 1.5 0.151 Width of paramere hook Not measured 0.04 (Æ0.016) 0.04 (Æ0.014) 0.03–0.05 0.03–0.05 À0.5 0.616

P values in bold indicate signiﬁcant differences.

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 6143 Cryptic Species in Biological Control I. D. Paterson et al.

Table 2. Morphological measurements of female Eccritotarsus catarinensis from two source populations.

Parameter Brazil (ÆSE), mm Peru (ÆSE), mm Brazil range, mm Peru range, mm t Value P value

Body length 2.95 (Æ0.031) 2.83 (Æ0.026) 2.79–3.17 2.73–2.92 À3.6 0.007 Body width 0.73 (Æ0.009) 0.66 (Æ0.010) 0.70–0.79 0.61–0.69 À5.6 <0.001 Pronotum length 0.44 (Æ0.023) 0.54 (Æ0.010) 0.32–0.50 0.38–0.48 À1.6 0.128 Pronotum width 0.83 (Æ0.010) 0.42 (Æ0.009) 0.79–0.88 0.73–0.83 À3.2 0.004 Head length 0.26 (Æ0.006) 0.26 (Æ0.010) 0.23–0.29 0.22–0.29 0.2 0.861 Head width 0.57 (Æ0.006) 0.54 (Æ0.005) 0.54–0.59 0.51–0.56 À3.1 0.006 Head vertex 0.30 (Æ0.008) 0.30 (Æ0.009) 0.27–0.36 0.26–0.33 À0.1 0.987 Antennae Segment 1 0.26 (Æ0.008) 0.3 (Æ0.005) 0.24–0.29 0.22–0.27 À2 0.059 Antennae Segment 2 0.45 (Æ0.005) 0.24 (Æ0.011) 0.44–0.55 0.41–0.53 À1 0.313 Antennae Segment 3 0.67 (Æ0.010) 0.48 (Æ0.010) 0.62–0.71 0.61–0.71 À0.2 0.871 Antennae Segment 4 0.60 (Æ0.009) 0.66 (Æ0.018) 0.46–0.70 0.50–0.66 À1.8 0.091

P values in bold indicate signiﬁcant differences.

Brazil♀ 9 Brazil♂ crosses produced offspring. Only four treatments (Brazil♀ 9 Brazil♂; Peru♀ 9 Peru♂) produced of the 24 replicates (17%) of Brazil♀ 9 Peru♂ crosses significantly more offspring than both interpopulation produced any offspring and no offspring were produced treatments (Brazil♀ 9 Peru♂; Peru♀ 9 Brazil♂). Multiple by Peru♀ 9 Brazil♂ crosses. comparisons of means revealed no significant difference Crosses of Peru♀ 9 Peru♂ produced an average of 32.9 between the two within-population treatments. nymphs (SE Æ 3.06) and crosses of Brazil♀ 9 Brazil♂ pro- A total of 16 hybrid individuals were produced, 12 of duced an average of 26.9 nymphs (SE Æ 5.33). An average which were analyzed using genetic techniques to confirm of only 0.67 nymphs (SE Æ 0.34) was produced by that they were true hybrids (see Genetic analysis for Brazil♀ 9 Peru♂ crosses and no offspring resulted from genetics methods). All hybrid individuals from the the reciprocal cross of Peru♀ 9 Brazil♂ (Fig. 1). There was Brazil♀ 9 Peru♂ crosses shared the identical Brazilian an overall significant difference between all treatments COI sequence from Taylor et al. (2011) and formed a (H = 54.05; P < 0.00001), where the within-population third group intermediate to the Brazilian and Peruvian parents in the nMDS from the ISSR data (Fig. 2).

Choice interbreeding experiment Two clearly defined groups were formed in the nMDS scatterplot from the ISSR data with all the Brazilian controls falling into Group 1 and all the Peruvian controls into Group 2 (Fig. 3). All the individuals that formed Group 1 had identical COI sequences to the Brazilian COI sequences from Taylor et al. (2011) and all those in Group 2 had identical sequences to the Peruvian sequences from the same study (Genbank accession numbers: Brazilian KU530108; Peruvian KU530109). No individuals were intermediate between the two groups and all individuals within a group had identical COI sequences (Fig. 3). This indicates that no hybrids were sampled in this experiment. When hybrid individuals from the no- choice experiment were included in this analysis, they Figure 1. The median number of nymphs produced by breeding pairs formed an intermediate group to the Brazilian and Peru- for each interbreeding treatment under no-choice conditions. Error vian groups in the nMDS plot similar to the hybrid group – bars represent the 25 75% range and lower case letters represent in Figure 2. significant differences according to a Multiple Comparison of Mean Ranks test (P < 0.05) with the same letter representing no significant Of the 50 individuals sampled from mixed populations, difference and different letters indicating where significant difference only nine grouped with the Peruvian samples (Group 2) occurred. B indicates Brazilian population, P indicates the Peruvian and 41 grouped with the Brazilian samples (Group 1) population. (Fig. 3). Peruvian individuals were sampled in three of

6144 ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. I. D. Paterson et al. Cryptic Species in Biological Control

Figure 2. Non-metric Multi-Dimensional Scaling (nMDS) scatterplot using the ISSR data for hybrid offspring and the parents of hybrid offspring. Hybrids share an identical COI proﬁle to their mothers and are intermediate to their parents in the MDS plot. B♀ indicates Brazilian mothers, P♂ the Peruvian fathers and H the hybrid offspring.

Figure 3. Non-metric Multi-Dimensional Scaling (nMDS) scatterplot using the ISSR data from the choice interbreeding experiment. ‘P’ and ‘B’ indicate individuals from the Brazilian and Peruvian control populations. Individuals from the six mixed populations are labeled 1– 6. For the COI region, all individuals in Group 1 were Brazilian, all individuals in Group 2 were Peruvian. the six populations (one in Pop. 1, six in Pop. 2 and two isolated, but are so similar morphologically that they were in Pop. 3) whereas all six populations had Brazilian indi- previously considered representatives of a single species. viduals (Fig. 3). As a consequence of our results, their morphological characteristics have been re-appraised and a formal taxonomic description is being prepared to separate the cryp- Screening for Wolbachia tic species from Peru as a new species of Eccritotarsus. The Wolbachia control produced clear bands of about The new species can still be considered as cryptic because 600 bp as was expected for insects infected with the Wol- the subtle morphological differences that have been iden- bachia bacteria. No bands were obtained from any tified would previously have been considered as indicative E. catarinensis individuals of either the Brazilian or Peru- of intraspecific variation. vian population, suggesting that neither of the popula- One of the few other confirmed cryptic species is the tions was infected by Wolbachia. sweet potato whitefly for which as many as 24 cryptic species were identified using genetic techniques (Dinsdale Discussion et al. 2010), many of which have been tested using crossbreeding experiments and confirmed as reproductively Cryptic species can be recognized and delimited according isolated species (Xu et al. 2010; Liu et al. 2012). There to the biological species concept using various methods are many cases where possible cryptic species have been and types of data, but this study is one of the few that identified although DNA-barcoding (Wilcox et al. 1997; has evaluated levels of infertility and reproductive incom- Gomez et al. 2002; Hebert et al. 2004a,b; Miura et al. patibility between cryptic insect taxa using a combination 2005; Ward et al. 2008; Saitoh et al. 2015) but because of molecular information and crossbreeding experiments. threshold percentage sequence divergence between species The two species within E. catarinensis are reproductively can differ dramatically between taxa, even within a single

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 6145 Cryptic Species in Biological Control I. D. Paterson et al. family, often overlapping with mean intraspecific diver- Province on the Yurapa River in northern Peru (Cordo gence (no ‘barcoding gap’; Meyer and Paulay 2005; Cog- 1999). No populations of E. catarinensis have been nato 2006; Park et al. 2011), these cannot be confidently recorded between southern Brazil and northern Peru but confirmed as true cryptic species without verifying the it is possible that populations do exist. The lack of gene levels of reproductive incompatibility between them. flow between the two populations may have resulted in Other problems associated with DNA-barcoding are the independent mutations that led to reproductive barriers presence of mitochondrial DNA sequences in the nuclear with limited morphological changes. Alternatively, the genome (numts) and heteroplasmy (more than one mito- genetic bottleneck that the Brazilian population experi- chondrial genome; Song et al. 2008). The use of alterna- enced after importation into South Africa, and subse- tive data sources, such as ISSR, morphology and mating quent population increase under relaxed selective experiments in this study, can act as checks on these pos- conditions, may have resulted in the formation of a new sible sources of error (Karanovic et al. 2016; Dejaco et al. species – a founder-flush effect (Carson 1971; Powell In press), as can more laborious methodological controls, 1978). After importation into quarantine in South Africa, although such practices compromise barcoding’s claim to all but one gravid female died (Taylor et al. 2011). In be a fast and efficient method of species discovery and addition, there has been no opportunity for subsequent delimitation (Song et al. 2008). gene flow with the source population, there is no evi- Hybrid E. catarinensis nymphs could only be produced dence of any bottlenecks in the source population, and by Brazilian females under no-choice conditions and the the founding population was introduced into a habitat extremely low number of offspring produced suggests that different from the source. These conditions are thought it is possible that no hybrid lineages would persist under to be required for founder effect or founder-flush specia- field conditions. The fact that no hybrids were sampled in tion (Giddings and Templeton 1983; Whitlock 1997; the choice experiment may be due to the fact that hybrids Fauergue et al. 2012) and have been present in supporting are extremely rare and were therefore not sampled, or it examples (Regan et al. 2003; Matute 2013; Oh et al. is possible that when mates from the same species were 2013). In addition, the population had human assistance available, no interbreeding between the two species in surviving the bottleneck, overcoming the large extinc- occurred. No-choice laboratory-based experiments have tion probability in extremely small populations. It is pos- consistently produced hybrids between well-established sible that a genetic reorganization in that gravid female species when interbreeding will not occur in wild popula- and her descendants render mating incompatible with the tions (Liu et al. 2012), so the hybrid nymphs produced in Peruvian, as well as Brazilian individuals in the source this study are very likely to be artefacts of the no-choice population. This hypothesis predicts reproductive isola- laboratory conditions under which the experiments were tion between the South African population derived from conducted. It is also possible that the small number of Brazil and that population in the native range. This could hybrid offspring produced in this study were sterile be tested by collecting E. catarinensis from the original hybrids and could therefore not persist or increase in collection site of the Brazilian population in Florianopo- abundance over time (Coyne and Orr 1989). Hybrid lis, Santa Catarina, and conducting interbreeding experi- sterility could not be tested in this study due to the low ments between that population and individuals from the number of hybrids that were produced and because two species in South Africa. The founder effect and foun- hybrids were killed to confirm their status using genetic der-flush models of speciation have received criticism methods. Hybrid sterility is believed to play an important (Barton and Charlesworth 1984; Coyne and Orr 2004) as role in the evolution of new species from allopatric popu- well as support (Slatkin 1996; Templeton 2008; Wessel lations (Orr 1995) and may therefore have contributed to et al. 2013; Pierce et al. 2016), so further study to deter- divergence in the E. catarinensis system. Similarly, the lack mine whether these theories have played a role in the of any hybrids produced by Peruvian females under both evolution of Eccritotarsus is warranted. choice and no-choice conditions may be due to complete Few biological control agents for the control of invasive incompatibility between Peruvian females and Brazilian alien plants are screened using genetic techniques prior to males, or it is possible that hybrids are so rare that none release but there are some records of cryptic species being were sampled in either experiment. discovered in biological control agent populations or in Little or no gene flow between E. catarinensis popula- populations of insects being considered as potential tions in Peru and Brazil would be expected due to the agents (Tosveski et al. 2013). Morphologically identical large geographic distance that separates the two popula- populations of the flea beetle, Psylliodes chalcomerus (Illi- tions. Eccritotarsus catarinensis has been recorded in the ger) [Coleoptera: Chrysomelidae], a potential biological states of Santa Catarina (Carvalho 1948) and Rio de control agent for yellow starthistle, Centaurea solstitialis L. Janeiro (Hill et al. 1999) in southern Brazil and in Loreto [Asteraceae], having a variety of different host plants in

6146 ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. I. D. Paterson et al. Cryptic Species in Biological Control the indigenous range, were all regarded as a single species appear to have slightly different host ranges indicates that but were divided into three possible cryptic species using the practice of importing multiple consignments of bio- genetic analyses (Gaskin et al. 2011). Tosveski et al. logical control agents for the purpose of increasing (2013) also identified cryptic species within the indige- genetic diversity of agent populations should be con- nous range of the Mecinus janthinus Germar [Coleoptera: ducted with extreme caution (Paynter et al. 2008). All Curculionidae] species complex, a weevil introduced into consignments, especially those collected from allopatric North America as a biological control agent for two spe- populations, should be treated as distinct entities with cies of toadflax (Linaria spp. [Scrophulariaceae]). potentially different host ranges until proven otherwise. Whether the cryptic species of M. janthinus are present in Different populations of the same species may have differ- the introduced populations in North America has not ent host ranges due to local adaptations (Karban 1989, been confirmed (Tosveski et al. 2013). A possible cryptic 1992; Hanks and Denno 1994) and in extreme cases, each species of the flea beetle, Aphthona lacertosa Rosenheim population may be a cryptic species. All consignments [Coleoptera: Chrysomelidae], a biological control agent should also be screened for the existence of cryptic species for leafy spurge, Euphorbia esula L. [Euphorbiaceae], was using basic genetic analyses such as sequencing the COI identified using genetic techniques after its release in region of a subset of consignments. North America (Roehrdanz et al. 2011). The leafy spurge Because both species occupy a similar ecological niche, flea beetle is the only example of a cryptic species of weed there is the possibility of competition between the two biological control agent that was discovered after release species if both are released as biological control agents. besides E. catarinensis. From the choice interbreeding experiment in this study, it The existence of cryptic species in biological control was clear that the Brazilian species had become more agent populations could result in unpredicted nontarget abundant than the Peruvian species because the majority effects if consignments of the agents that are exported of the samples taken were Brazilian. This suggests that the from the native range are assumed to have identical host Brazilian species was a better competitor under those con- ranges. Paynter et al. (2008) discovered that two popula- ditions. Temperature is likely to be an important factor tions of the gorse pod moth, Cydia succedana (Denis and in determining which of the two species is a stronger Schiffermuller)€ [Lepidoptera: Olethreutidae], a biological competitor because the Peruvian species was collected control agent for the invasive alien plant gorse, Ulex euro- from a much warmer region than the Brazilian. The Peru- paeus L. [Fabaceae], in New Zealand, differed in their vian species should therefore be released at sites in warm, host ranges. The two populations were sourced from geo- tropical regions and the Brazilian species in cooler more graphically isolated populations in the native range of the temperate regions. Further study of the thermal physiolo- moth but the low sequence divergence indicated that the gies of both species are, however, required. two populations were not cryptic species (Paynter et al. The mechanisms that cause reproductive isolation 2008). With a higher sequence divergence between the between the two species have not been identified. The two Eccritotarsus species, there is a higher likelihood that genitalia, and more specifically the parameres, of Eccrito- the host ranges of the species could differ (Taylor et al. tarsus are used as characters to define the various species 2011) and host specificity data indicate that there are sub- in the genus (Carvalho 1948) and there are only relatively tle differences in the host ranges of the two species (I.D. small differences in these features between the two species Paterson, unpublished data). Fortunately, both popula- but the small differences may be sufficient to prevent tions were subjected to host specificity testing before sperm transfer. The bacterium Wolbachia has been ruled being released in South Africa despite the fact that host out as a cause of the reproductive isolation in this study specificity testing of the Peruvian consignment was tech- because it is not present in either species. Preliminary nically not required because the species had already been studies suggest that premating behavioral barriers may be cleared for release (I.D. Paterson, unpublished data). Both involved and studies are now underway to determine species are safe for release in South Africa and no nontar- rates of copulation events between the two species under get effects are predicted because, although there are subtle choice and no-choice conditions, whether sperm transfer differences in the host ranges, these differences are is successful for between-species matings, and whether restricted to close relatives of the target plant that are not pheromones may play a role in reproductive isolation. indigenous or grown commercially in South Africa. Speciation has occurred but the speciation has resulted in Although in this case the two consignments of the agent very limited morphological change. The changes that have were both subjected to host specificity testing, this is not resulted in reproductive isolation of the two species may necessarily common practice in all biological control pro- therefore be present in some other characteristic that grammes. The fact that the two consignments of the agent could be physical or behavioral (Bickford et al. 2007).

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 6147 Cryptic Species in Biological Control I. D. Paterson et al.

Acknowledgments biological control of water hyacinth, Eichhornia crassipes. Aquat. Bot. 86:179–186. Part of the funding for this work was provided by the Coetzee, J. A., M. P. Hill, M. P. Byrne, and A. Bownes. 2011. South African Research Chairs Initiative of the Depart- A review of the biological control programmes on ment of Science and Technology and the National Eichhornia crassipes (C. Mart.) Solms (Pontederiaceae), Research Foundation of South Africa. Any opinion, find- Salvinia molesta D.S. Mitch. (Salviniaceae), Pistia stratiotes ing, conclusion or recommendation expressed in this L. (Araceae), Myriophyllum aquaticum (Vell.) Verdc. material is that of the authors and the NRF does not (Haloragaceae) and Azolla filiculoides Lam. (Azollaceae) in accept any liability in this regard. Funding was also pro- South Africa since 1999. Afr. Entomol. 19:451–468. vided by the Working for Water (WfW) programme of Cognato, A. I. 2006. Standard percent DNA sequence the Department of Environmental Affairs: Natural difference for insects does not predict species boundaries. J. – Resource Management programme (DEA:NRM). Econ. Entomol. 99:1037 1045. Colautti, R. I., M. Manca, M. Vijanen, H. A. M. Ketelaars, H. Burgi,€ H. J. Macisaac, et al. 2005. Invasion genetics of the Conflict of Interest Eurasian spiny waterflea: evidence for bottlenecks and gene flow using microsatellites. Mol. Ecol. 14:1869–1879. None declared. Cook, L. G., R. D. Edwards, M. D. Crisp, and N. B. Hardy. 2010. Need morphology always be required for new species References descriptions? Invertebr. Syst. 24:322–326. Abad-Franch, F., M. G. Pavan, N. Jaramillo-O, F. S. Cordo, H. A. 1999. New agents for biological control of water Palomeque, C. Dale, D. Chaverra, et al. 2013. Rhodnius hyacinth. Pp. 68–74 in M. P. Hill, M. H. Julien and T. D. barretti, a new species of Triatominae (Hemiptera: Center, eds. Proceedings of the first IOBC Global Working Reduviidae) from western Amazonia. Mem. Inst. Oswaldo Group Meeting for the Biological and Integrated Control of Cruz 108(Suppl. 1):92–99. Water Hyacinth, 16–19 November 1998, St Lucia Park Abbot, P. 2001. Individual and population variation in Hotel, Harare, Zimbabwe. Plant Protection Research invertebrates revealed by Inter-simple Sequence Repeats Institute, Pretoria. (ISSRs). J. Insect Sci. 1:1–3. Coyne, J. A., and H. A. Orr. 1989. Patterns of speciation in Arrigo, N., J. W. Tuszynski, D. Ehrich, T. Gerdes, and N. Drosophila. Evolution 43:362–381. Alvarez. 2009. Evaluating the impact of scoring parameters Coyne, J. A., and H. A. Orr. 2004. Speciation. Sinauer on the structure of intra-specific genetic variation using Associates, Sunderland, MA545 Pp. RawGeno, an R package for automating AFLP scoring. BMC Darlington, C. D. 1940. Taxonomic systems and genetic Bioinformatics 23:34–56. systems. Pp. 137–160 in J. Huxley, ed. The new systematics. Barton, N., and B. Charlesworth. 1984. Genetic revolutions, Clarendon Press, Oxford, U.K. founder effects, and speciation. Annu. Rev. Ecol. Syst. Davis, M. A., M. R. Douglas, M. L. Collyer, and M. E. 15:133–164. Douglas. 2016. Deconstructing a species-complex: geometric Bickford, D., D. J. Lohman, S. S. Navjot, P. K. L. Ng, R. morphometric and molecular analyses define species in the Meier, R. Winker, et al. 2007. Cryptic species as a window Western Rattlesnake (Crotalus viridis). PLoS One 11: on diversity and conservation. Trends Ecol. Evol. 22:148– e0146166. 155. Dejaco, T., M. Gassner, W. Arthofer, B. C. Schlick-Steiner, and Bordenstein, S. R., F. P. O’Hara, and J. H. Werren. 2001. F. M. Steiner. In press. Taxonomist’s nightmare ... Wolbachia-induced incompatibility precedes other hybid evolutionist’s delight: an integrative approach resolves incompatibilities in Nasonia. Nature 409:707–710. species limits in jumping bristletails despite widespread Carson, H. L. 1971. Speciation and the founder principle. hybridization and parthenogenesis. Syst. Biol. Stadler Genet. Symp. Univ. Missouri 3:51–70. Dinsdale, A., L. Cook, C. Riginos, Y. M. Buckley, and P. J. de Carvalho, J. C. M. 1948. Mırideos neotropical, XXXI: Barro. 2010. Refined global analysis of Bemisia tabaci Gênero “Pseudobryocoris” Distant e uma especie nova de (Gennadius) (Hemiptera: Sternorrhyncha: Aleyrodidae) “Neella” Reuter (Hemiptera). An. Acad. Bras. Ciênc. mitochondrial COI to identify species level genetic 54:95–104. boundaries. Ann. Entomol. Soc. Am. 103:196–208. Coetzee, J. A., M. J. Byrne, and M. P. Hill. 2007a. Predicting Fauergue, X., E. Vercken, T. Malausa, and R. A. Hufbauer. the distribution of Eccritotarsus catarinensis, a natural enemy 2012. The biology of small, introduced populations, with released on water hyacinth in South Africa. Entomol. Exp. special reference to biological control. Evol. Appl. 5:424– Appl. 125:237–247. 443. Coetzee, J. A., M. J. Byrne, and M. P. Hill. 2007b. Impacts of Gaskin, J. F., M. C. Bon, M. J. W. Cock, M. Cristofaro, A. De nutrients and herbivory by Eccritotarsus catarinensis on the Biase, R. De Clerck-Floate, et al. 2011. Applying molecular-

6148 ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. I. D. Paterson et al. Cryptic Species in Biological Control

based approaches to classical biological control of weeds. Machtelinckx, T., T. Van Leeuwen, B. Vanholme, B. Biol. Control 58:1–21. Gehesquiere, W. Dermauw, B. Vandekerkhove, et al. 2009. Giddings, L. V., and A. R. Templeton. 1983. Behavioral Wolbachia induces strong cytoplasmic incompatibility in the phylogenies and the direction of evolution. Science 220:372– predatory bug Macrolophus pygmaeus. Insect Mol. Biol. 378. 18:373–381. Gomez, A., M. Serra, G. R. Carvalho, and D. H. Lunt. 2002. Matute, D. R. 2013. The role of founder effects on the evolution Speciation in ancient cryptic species complexes: evidence of reproductive isolation. J. Evol. Biol. 26:2299–2311. from the molecular phylogeny of Brachionus plicatilis Mayr, E. 1942. Systematics and the origin of species. Columbia (Rotifera). Evolution 56:1431–1444. Univ. Press, New York. Hammer, Ø., and D. A. T. Harper. 2001. PAST: Mayr, E. 1963. Animal species and evolution. The Belknap paleontological statistics software package for education and Press of Harvard Univ. Press, Cambridge, U.K. data analysis. Paleontol. Electronica 4:1–9 (Art 4). Meyer, C. P., and G. Paulay. 2005. DNA barcoding: error rates Hanks, L. M., and R. F. Denno. 1994. Local adaptation in the based on comprehensive sampling. PLoS Biol. 3:e422. armored scale insect Pseudaulacaspis pentagona (Homoptera: Miura, O., A. M. Kuris, M. E. Torchin, R. F. Hechinger, E. J. Diaspididae). Ecology 75:2301–2310. Dunham, and S. Chiba. 2005. Molecular-genetic analysis Hebert, P. D. N., A. Cywinski, S. L. Ball, and J. R. de Waard. reveals cryptic species trematodes of intertidal gastropod, 2003. Biological identifications through DNA barcodes. Batillaria cumingi (Crosse). Int. J. Parasitol. 35:793–801. Proc. R. Soc. Lond. B Biol. Sci. 270:313–321. Noor, M. A. F. 2002. Is the biological species concept showing Hebert, P. D. N., G. H. Penton, J. M. Burns, D. H. Janzen, its age? Trends Ecol. Evol. 17:153–154. and W. Hallwachs. 2004a. Ten species in one: DNA Oh, K. P., G. L. Conte, and K. L. Shaw. 2013. Founder effects barcoding reveals cryptic species in the Neotropical skipper and the evolution of asymmetrical sexual isolation in a butterfly Astraptes fulgerator. PNAS 101:14812–14817. rapidly-speciating clade. Curr. Zool. 59:230–238. Hebert, P. D., M. Y. Stoeckle, T. S. Zemlak, and C. M. Francis. Orr, H. A. 1995. The population genetics of speciation: the 2004b. Identification of birds through DNA barcodes. PLoS evolution of hybrid incompatibilities. Genetics 139:1805– Biol. 2:E312. 1813. Hill, M. P., C. J. Cilliers, and S. Neser. 1999. Life history and Park, D. S., R. Foottit, E. Maw, and P. D. N. Hebert. 2011. laboratory host range of Eccritotarsus catarinensis (Carvalho) Barcoding bugs: DNA-based identification of the true bugs (Hemiptera: Miridae), a new natural enemy released on (Insecta: Hemiptera: Heteroptera). PLoS One 6:e18749. water hyacinth (Eichhornia crassipes (Mart.) Solms-Laub.) Paynter, Q., A. H. Gourlay, P. T. Oboyski, S. V. Fowler, R. L. (Pontederiaceae) in South Africa. Biol. Control 14:127–133. Hill, T. M. Withers, et al. 2008. Why did specificity testing Kaiser, L., B. P. Le Ru, F. Kaoula, C. Paillusson, C. fail to predict the field host-range of the gorse pod moth in Capdevielle-Dulac, J. O. Obonyo, et al. 2015. Ongoing New Zealand? Biol. Control 46:453–462. ecological speciation in Cotesia sesamiae, a biological control Pierce, A. A., M. P. Zalucki, M. Bangura, M. Udawatta, M. R. agent of cereal stem borers. Evol. Appl. 8:807–820. Kronforst, S. Altizer, et al. 2016. Serial founder effects and Karanovic, T., M. Djurakic, and S. M. Eberhard. 2016. Cryptic genetic differentiation during worldwide range expansion of species or inadequate taxonomy? Implementation of 2D monarch butterflies. Proc. R. Soc. Lond. B Biol. Sci. geometric morphometrics based on integumental organs as 281:20142230. landmarks for delimitation and description of copepod taxa. Powell, J. R. 1978. The founder-flush speciation theory: an Syst. Biol. 65:304–327. experimental approach. Evolution 32:465–474. Karban, R. 1989. Fine-scale adaptation of herbivorous thrips to Regan, J. L., L. M. Meffert, and E. H. Bryant. 2003. A direct individual host plants. Nature 340:60–61. experimental test of founder-flush effects on the evolutionary Karban, R. 1992. Plant variation: its effects on populations of potential for assortative mating. J. Evol. Biol. 16:302–312. herbivorous insects. Pp. 195–215 in R. S. Fritz and E. L. Rissler, L. J., and J. J. Apodaca. 2007. Adding more ecology Simms, eds. Plant resistance to herbivores and pathogens. into species delimitation: ecological niche models and Ecology, evolution and genetics. The Univ. of Chicago Press, phylogeography help define cryptic species in the Black Chicago and London. Salamander (Aneides flavipunctatus). Syst. Biol. 56:924–942. Lachmuth, S., W. Durka, and F. M. Schurr. 2010. The Roehrdanz, R., R. Bourchier, A. Cortilet, D. Olson, and S. making of a rapid plant invader: genetic diversity and Sears. 2011. Phylogeny and genetic diversity of flea beetles differentiation in the native and invaded range of Senecio (Aphthona sp.) introduced to North America as biological inaequidens. Mol. Ecol. 19:3952–3967. control agents for leafy spurge. Ann. Entomol. Soc. Am. Liu, S. S., J. Colvin, and P. J. De Barro. 2012. Species concepts 104:966–975. as applied to the whitefly Bemisia tabaci systematics: how Ross, K. G., and D. D. Shoemaker. 2005. Species delimitation in many species are there? J. Integr. Agric. 11:176–186. native South American fire ants. Mol. Ecol. 14:3419–3438.

ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd. 6149 Cryptic Species in Biological Control I. D. Paterson et al.

Saitoh, T., N. Sugita, S. Someya, Y. Iwami, S. Kobayashi, H. species within Mecinus janthinus species complex (Mecinini, Kamigaichi, et al. 2015. DNA barcoding reveals 24 distinct Curculionidae) introduced in North America for biological lineages as cryptic bird species candidates in and around the control of invasive toadflaxes. Biocontrol 58:563–573. Japanese Archipelago. Mol. Ecol. 15:177–186. Ward, R. D., B. H. Holmes, and G. K. Yearsley. 2008. DNA Schlick-Steiner, B. C., B. Seifert, C. Stauffer, E. Christian, R. barcoding reveals a likely second species of Asian sea bass H. Crozier, and F. M. Steiner. 2007. Without morphology, (barramundi) (Lates calcarifer). J. Fish Biol. 72:458–463. cryptic species stay in taxonomic crypsis following Wessel, A., H. Hoch, M. Asche, T. von Rintelen, B. Stelbrink, discovery. Trends Ecol. Evol. 22:391–392. V. Heck, et al. 2013. Founder effects initiated rapid species Simoes,~ P. M., G. Mialdea, D. Reiss, M.-F. Sagot, and S. radiation in Hawaiian cave planthoppers. Proc. Natl Acad. Charlat. 2011. Wolbachia detection: an assessment of Sci. USA 110:9391–9396. standard PCR protocols. Mol. Ecol. Resour. 11:567–572. Whitlock, M. C. 1997. Founder effects and peak shifts without Slatkin, M. 1996. In defense of founder-flush theories of genetic drift: adaptive peak shifts occur easily when speciation. Am. Nat. 147:493–505. environments fluctuate slightly. Evolution 51:1044–1048. Song, H., J. E. Buhay, M. F. Whiting, and K. A. Crandall. Wilcox, T. P., L. Hugg, J. A. Zeh, and D. W. Zeh. 1997. 2008. Many species in one: DNA barcoding overestimates Mitochondrial DNA sequencing reveals extreme genetic the number of species when nuclear mitochondrial differentiation in a cryptic species complex of Neotropical pseudogenes are coamplified. Proc. Natl Acad. Sci. USA pseudoscorpions. Mol. Phylogenet. Evol. 7:208–216. 105:13486–13491. Winston, R. L., M. Schwarzl€ander, H. L. Hinz, M. D. Day, M. Souza, N. A., C. A. Andrade-Coelho, F. M. Vigoder, R. D. J. Cock, and M. H. Julien. 2014. Biological control of weeds: Ward, and A. A. Peixoto. 2008. Reproductive isolation a world catalogue of agents and their target weeds. 5th ed. between sympatric and allopatric Brazilian populations of USDA Forest Service, Forest Health Technology Enterprise Lutzomyia longipalpis s.l. (Diptera: Psychodidae). Mem. Inst. Team, Morgantown, West Virginia. FHTET-2014-04. 838 Oswaldo Cruz 103:216–219. Pp. Taylor, S. J., D. A. Downie, and I. D. Paterson. 2011. Genetic Xu, J., P. J. de Barro, and S. S. Liu. 2010. Reproductive diversity of introduced populations of the water hyacinth incompatibility among genetic groups of Bemisia tabaci biological control agent Eccritotarsus catarinensis supports the proposition that the whitefly is a cryptic (Hemiptera: Miridae). Biol. Control 58:330–336. species complex. Bull. Entomol. Res. 100:359–366. Templeton, A. R. 2008. The reality and importance of founder Zhou, W., F. Rousset, and S. O’Neil. 1998. Phylogeny and speciation in evolution. BioEssays 30:470–479. PCR-based classification of Wolbachia strains using wsp Tosveski, I., J. Jovic, O. Krstic, and A. Gassmann. 2013. PCR- gene sequences. Proc. R. Soc. Lond. B Biol. Sci. 265:509– RFLP-based method for reliable discrimination of cryptic 515.

6150 ª 2016 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.